Eh, ctroencephalography and clinical Neurophysiologv, 1991, 78:361-377
' 1991 Elsevier Scientific Publishers Ireland, Ltd. 0013-4649/91/$03.50
Event-related potentials and eyeblink responses in automatic and controlled
processing: effects of age
Judith M. Ford and Adolf Pfefferbaum
Department of Ps3'chiattT and Behaeioral Sciences, Stanford Universi(v School of Medicine. and Palo A ho VA Medical Center. Palo Alto, ('A (U. S.A.)
(Accepted for publication: 10 May 1990)
potential (ERP) paradigms designed to elicit responses in reaction time tasks and to a startling noise burst. EFG was analyzed from 17 standard
10 20 electrode sites. Reaction time and performance data suggested that the elderly did not perform worse than the young. Nevertheless, the
physiological responses of the elderly differed significantly from those of the young. While the task-dependent P3s at Pz were smaller in the elderly
than in the young, the automatic P3 was smaller yet. The distribution of both types of P3 across the scalp was more uniform in the elderly than in
the young. Single-trial analyses revealed that the P3 amplitude differences at Pz were~not due to latency dispersal of single trials. Single-trial startle
eye blink responses to intense noise bursts during the automatic paradigm were considerably less frequent in the elderly, although their individual
startle blinks were actually larger. The data demonstrate that the electrophysiological responses of the elderly are different from the young both in
tasks eliciting automatic responses and in tasks requiring controlled processing.
Seventeen young (mean age - 20.2 years old) and 16 elderly (mean age = 72.6 years old) women were tested with event-related
Key words: Event-related potential; Eyeblink response: Automatic processing: Controlled processing: P3; Reaction time task: Age-related response
Hasher and Zacks (1979) have proposed a distinction
between automatic and controlled cognitive mecha-
nisms. Automatic operations drain very little from a
hypothetical, limited-capacity processor, do not inter-
fere with other ongoing cognitive activity+ and have
been shown not to be affected by age (Hasher and
Zacks 1979). The ability to estimate frequency of occur-
rence is considered automatic and is spared in the
elderly (Hasher and Zacks 1979), as is probability
estimation (Ford et al. 1982a). Controlled processes, in
contrast, take considerable effort, benefit from practice,
involve rehearsal, may involve elaborative mnemonic
activities+ and are initiated intentionally. A body of
literature exists showing that the elderly are deficient on
tasks requiring controlled processes (see Rabbitt 1981).
While the distinction between controlled and auto-
matic processing has been well elaborated, it is possible
to consider some overlap between the two. For example,
controlled processing might be initiated automatically,
: This investigation was supported by the Medical Research Service of
the Department of Veterans Affairs, MH40052, MH30854, MH40041.
Correspondence to: Dr. J.M. Ford+ Ph.D., M.D., Dept. of Psychi-
atry, Stanford University of Medicine, Palo Alto VA Medical Center.
3801 Miranda Avenue, Palo Alto, CA 94304 (U.S.A.).
or be preceded by an automatic stage. Ohman (1979)
suggested that the orienting response has two stages+ the
preattentive (i.e., automatic) processing of a stimulus
followed by more complicated, selective, resource-
limited (i.e., controlled) processing.
In tasks requiring active involvement of the subject,
the amplitude of P3 of the event-related potential (ERP)
has been observed to be reduced (Marsh 1975; Goodin
et al. 1978; Picton et al. 1984: Mullis et al. 1985) or
changed in scalp distribution in the elderly (Marsh
1975; Ford et al. 1979b, 1982b: Pfefferbaum et al.
1980a,b, 1984; Picton et al. 1984: Mullis et al. 1985;
Looren de Jong et al. 1988). Not only does P3 ampli-
tude and distribution change with age, P3 latency is also
delayed. This has generally been observed using an
auditory or visual oddball paradigm in which subjects
count or press a response key to infrequent target
stimuli. (See Ford and Pfefferbaum (1985) for a review.)
While the P3 component is usually elicited in situa-
tions believed to require controlled processing, Roth et
al. (1984) have developed a paradigm that elicits P3 to
an intense noise burst in the absence of a stimulus-re-
lated task. Because this P3 is elicited even when atten-
tion is directed away, i.e., to a demanding visual track-
ing task, it is considered to be principally automatic
(Roth et al. 1984). Though this P3 is principally auto-
matic, the fact that its amplitude can be enhanced by
task requirements indicates that the process, once ini-
362 J.M. FORD, A. PFEFFERBAUM
tiated, is to some extent resource-limited. We will call
this the automatic P3, and those elicited in the RT
paradigms the task dependent P3s.
This automatic P3 paradigm also elicits a large eye
blink response, which we consider to be a component of
the automatically elicited startle response (Pfefferbaum
et al. 1989). The auditory startle response is elicited
automatically, is a simple response with a well worked
out neural circuit, and has a number of special ad-
vantages for human research (Geyer and Braff 1987).
The N1 component of the ERP may provide an
additional measure of automatic responses. While some
portion of the negativity associated with N1 may be
affected by controlled processes, like selective attention
or general attention (Hansen and Hillyard 1980),
another portion is associated with the more involuntary
automatic processing of the signal (Na~itanen and Pic-
Most studies of ERPs in the elderly have used the
traditional midline sites, Fz, Cz, and Pz. In young adult
subjects, P3 has a frontal minimum and a parietal
maximum, while in elderly, P3 has a more central scalp
distribution, being about equal in amplitude across the
3 midline sites. The lack of frontal slow wave negativity
in the elderly may be responsible for part (Picton et al.
1984) or most (Pfefferbaum et al. 1984) of the apparent
uniform anterior/posterior distribution of P3. Indeed,
several investigators have reported a reduction in fron-
tal slow wave negativity in the elderly (Michalewski et
al. 1980; Pfefferbaum et al. 1984; Looren de Jong et al.
1988). The use of additional scalp electrodes provides
an opportunity to determine the extent of this unifor-
mity of distribution across the entire scalp.
We now report on a study in which ERPs and startle
blink responses were used to assess age-related changes
in automatic and controlled processing. Multiple elec-
trode sites were used to better elaborate the topography
of the age-related changes in P3. In addition, single
trials were latency-adjusted to address the possibility
that latency dispersal was responsible for age-related P3
Seventeen healthy young (17-24 years, mean = 20.2
years) and 16 older (64-83 years, mean = 72.6 years)
female volunteers participated. Subjects were recruited
and screened by telephone interview and questionnaire
to exclude those with a history of significant psychiatric
or neurological histories, recent use of psychoactive
drugs, or alcohol consumption exceeding 50 g/day. One
elderly subject initially tested in this group was ex-
cluded post hoc (i.e,, after data collection, but before
group statistical analysis) when consultation with col-
leagues identified her as someone who had been re-
ferred to the hospital 2 years prior to ERP testing,
complaining of memory problems.
ERPs were collected during all 3 of the following
paradigms. Stimuli for all paradigms were generated
and data collected by a PDP-11/34 computer.
Subjects wore earphones and
sat upright in an easy chair in a sound-attenuated room.
They were presented with a series of 300 auditory
stimuli with a fixed interstimulus interval of 1.5 sec.
Frequent tones were 500 Hz, 80 dB SPL, 50 msec
duration and occurred on 80% of the trials. The other
20% (rares) were an intense white noise burst, 105 dB
SPL, 50 msec duration. Tone stimuli had a shaped rise
and fall time of 5 msec while noises rose and fell in 100
t~sec. Stimuli were presented in a Bernoulli sequence,
held constant across subjects. No task was assigned to
ters were identical to the automatic paradigm, except
that the stimulus which occurred 20% of the time was a
1000 Hz, 80 dB SPL, 50 msec duration tone. Subjects
were asked to press a reaction time (RT) button to the
rare stimuli as quickly as possible, giving equal impor-
tance to speed and accuracy.
Visual RT paradigm.
This was a visual version of
the auditory RT paradigm, in which stimuli appeared
on a CRT. Minus signs (-) appeared on 80% of the
trials, and plus signs (+) on the other 20%. The task
was the same as for the auditory RT paradigm.
Stimulus sequence parame-
EEG and EOG recordings
EEG was recorded from the 19 standard 10-20 sites
using an electrode cap. A sterno-vertebral reference was
used with a balancing circuit to minimize EKG artifacts
(Stephenson and Gibbs 1951). Technical difficulties with
the occipital electrodes in many of the subjects necessi-
tated eliminating those wave forms from the data analy-
sis, leaving data from 17 scalp electrode sites and 2 eye
derivations for analysis. Vertical EOG was recorded
from electrodes placed above and below the right eye,
and horizontal EOG from electrodes placed at the outer
canthi of each eye, EEG and EOG were sampled every
5 msec from 100 msec prior to stimulus onset to 1150
msec post stimulus. A microcomputer-controlled Ni-
hon-Kohden EEG-4221 system (Tokyo, Japan) re-
corded EEG at a gain of 10 K and EOG at 2.7 K with a
bandpass of 0,13-70 Hz.
ERP data analysis
The single-trial ERPs were edited to exclude those
associated with an incorrect behavioral response and
electrical artifacts (> +//-200 /,V). Trials con-
ERPs AND EYEBLINK RESPONSES IN AUTOMATIC AND CONTROLLED PROCESSING 363
taminated with VEOG or HEOG artifact (> +/-
t~V) were rejected from analysis of the auditory and
visual RT tasks. Because the blink reflex is an integral
part of the response to the intense noise, trials contain-
ing startle blink responses (before 275 msec) were not
rejected from the analysis of the automatic paradigm,
although trials containing blinks between 275 and 600
msec post stimulus were rejected. A subject's data were
dropped from further analysis of all data for a particu-
lar paradigm if 15 or fewer rare trials remained follow-
ing these screening procedures. Two elderly and 3 young
were rejected from the automatic paradigm; 2 elderly
and 4 young were rejected from the auditory RT para-
digm, and 2 elderly and 2 young were rejected from the
visual RT paradigm. One elderly subject was excluded
from all 3 paradigms due to a failure of accurate ampli-
fier calibration, and another was excluded due to broken
electrodes at T5 and T6. The number of elderly and
young subjects remaining were 11 and 14 in the auto-
matic paradigm, 11 and 13 in the auditory RT para-
digm, and 11 and 15 in the visual RT paradigm.
Following this artifact- and error-rejection proce-
dure, 3 types of analysis were performed on the EEG
data, each of which is described in detail below: tradi-
tional analyses of peak amplitudes and latencies, scalp
distribution analyses using multi-lead data, and single-
trial analyses of P3.
In these analyses, separate
conventional ERP averages were derived for responses
to the rare and frequent stimuli from each paradigm.
Amplitude and latency measurements were done for
each component only at the lead where that component
is usually largest. For both rare and frequent event
ERPs, component peaks were identified as the maxi-
mum voltage in the following ranges: N1 at Cz (70-150
msec), P2 at Fz (N1-225 msec), and N2 at Fz (between
P2 and P3). For the rare event ERPs, P3 was measured
at Fz, Cz, and Pz between 275 and 600 msec; N4 at Fz
was measured as a negative peak maximum in the 200
msec following P3; and the slow wave at FP1 and FP2
was measured as the area between 400 and 700 msec.
To elucidate possible age-related effects of the slow
wave on P3 at these sites, P3 amplitude was also mea-
sured at FP1 and FP2. In analyses of more than one
electrode site, a repeated measures ANOVA was used.
Greenhouse-Geisser corrections were applied where ap-
propriate and are designated GG.
Scalp distribution of P3 analyses.
proaches were taken to describe age-related differences
in the scalp topography of P3.
(1) The amplitude variability across the 17 leads was
calculated for each subject, using a root mean squared
voltage deviation ('RMS') analysis (Lehmann and
Skrandies 1980). For each subject, a common average
reference was calculated by averaging the data, time
point by time point across all 17 leads. This average was
The following ap-
then subtracted from the activity at each individual
electrode and the root mean squared deviation across
the 17 leads was calculated for each time point. To
determine the significance of age-related differences in
the variability of the P3 amplitude across the 17 leads,
the RMS value at the P3 latency, determined at Pz, for
each subject was entered into separate t tests for each
paradigm. A similar analysis was performed on the N1
data to the frequent tones in the automatic and auditory
(2) P3 amplitude (at the Pz-P3 latency) was mea-
sured across a grid of 15 electrode sites. This allowed
analysis of 3 coronal rows from the 5 electrodes at each
of the frontal (F7, F3, Fz, F4, F8), central (T3, C3, Cz,
C4, T4) and parietal (T5, P3, Pz, P4, T6) locations, as
well as 5 sagittal rows from 3 electrodes each at the
far-left (F7, T3, T5), mid-left (F3, C3, P3), midline (Fz,
Cz, Pz), mid-right (F4, C4, P4), and far-right (F8, T4,
T6) locations. In addition to analyses of the non-trans-
formed P3 amplitudes at these electrode sites, the data
were also normalized according to the suggestion of
McCarthy and Wood (1985):
P3(i') = [P3(i) - P3(min)]/[P3(max) - P3(min)]
where i = a given electrode site and i' = the normalized
value, and where min = within-group minimum value of
15 electrode sites and max = within-group maximum
value of 15 electrode sites. Both normalized and non-
transformed data sets were subjected to ANOVAs for
the repeated factors of laterality (5 levels) and anterior/
posterior (3 levels). Post hoc pairwise comparisons were
made if indicated.
This procedure, originally de-
scribed in Pfefferbaum and Ford (1988), includes im-
provements to enhance the goodness of fit of the tem-
plate and guard against bias introduced by low signal to
noise ratios. It was applied to rare event ERPs elicited
by the auditory RT and visual RT paradigms, but not
the automatic paradigm due to startle eye blink con-
First, using only trials that had passed the other
artifact- and error-rejection screens, the digitized EEG
recorded from Pz was reduced to a 10 msec sample rate
by omitting every other point, and a low pass digital
filter at 3.5 Hz (down 3 dB; Ruchkin and Glaser 1978)
was applied. A 2.0 Hz half sine wave was used as the
template. It was moved across each trial in 10 msec
increments over 2 latency ranges: 280-600 msec (signal
range) and 610-930 msec (noise range). The covariance
between the template and the corresponding points of
the single trials was calculated at each positioning. The
latency in the signal range at which the maximum
covariance was found was tentatively identified as P3
latency for that trial. The tentative P3 signal was only
accepted as a bona fide signal if it met 2 criteria: its
364 J,M. FORD, A. PFEFFERBAUM
covariance had to be larger than the maximum covari-
ance in the noise range on that trial, and its wave shape
had to correlate with the template at the P < 0.05 level.
If both conditions were not met, the trial was deemed to
have not passed the signal/noise screen and was ex-
cluded from this analysis. The remaining single-trial P3
latency estimates were used to calculate the median P3
latency value, correlations between P3 latency and RT
on individual trials, and to construct latency-adjusted as
well as stimulus-synchronized averages. A latency-ad-
justed average is one in which P3s on each individual
trial at each lead are all aligned at 250 msec (an
arbitrarily selected latency) before summation. This
procedure was applied to the data recorded from each
of the 17 scalp electrodes using the Pz-P3 latency for
The median P3 latencies, the stimulus-synchronized
P3 amplitudes, and the latency-adjusted P3 amplitudes
at Pz were compared across groups using t tests. Ad-
ditionally, the amplitudes derived from the stimulus-
synchronized and latency-adjusted techniques were
compared to each other across groups using repeated
measures ANOVAs. The numbers of trials not passing
the signal/noise screen were compared across groups.
The RTs on those trials were compared to RTs on trials
that did pass the signal/noise screen using t tests.
Group differences in estimates of goodness of fit be-
tween the template and the signal epoch (cross-products
and correlation coefficients) were assessed using t tests.
Startle blink analysis.
Initially, VEOG peak ampli-
tudes elicited during the automatic paradigm were mea-
sured from the average VEOG tracing, for each individ-
ual, constructed following the EEG artifact rejection
procedures. Next, single-trial analyses of the VEOG
data were performed in 2 ways, both independent of
EEG artifact rejection procedures. First, the numbers of
blinks were compared across groups using single-trial
analyses. Single-trial VEOG excursions exceeding + 100
/,V were counted as blinks. Startle blinks were consid-
ered as occurring between 50 and 150 msec immediately
following the loud noise; spontaneous blinks were con-
sidered as occurring between 0 and 1000 msec following
the frequent tone. Because of the possibility of the
startle blinks in the elderly falling beyond the short
latency window, an additional window out to 250 msec
post stimulus was also searched. The numbers of blinks
were compared across groups using t tests. Second, the
amplitude and latency of each individual startle blink
were measured. These values were averaged for an indi-
vidual subject and the average amplitudes and latencies
were compared across groups using t tests.
To determine whether VEOG amplitudes diminished
over the course of the experiment, single-trial VEOG
peak amplitudes were measured between 50 and 150
msec following the loud noise and between 0 and 1000
msec following the frequent tone. It is important to note
that because not all trials contained startle blinks, this is
only a rough estimate of response decrement over time,
especially in the elderly who blinked on very few of the
startle trials. Nevertheless, these VEOG amplitudes were
regressed against trial number for each group, using a
linear regression analysis. Because these responses were
not individually identified as blinks, they are referred to
as VEOG peak amplitudes in the text of this report.
VEOG responses individually identified as blinks are
referred to as blinks.
Conventional grand average wave forms of the ERPs
to the infrequent events in the 3 paradigms are dis-
played in Fig. la, b, and c, with the data from the 2 age
groups superimposed. Data from all 17 scalp electrodes
and the vertical and horizontal eye movement channels
are included in this figure,
VEOG in the automatic paradigm
As can be seen in the average VEOG tracing in Fig.
la, the average startle VEOG peak amplitude to the
intense noise burst was much smaller in the old subjects
compared to the young (t (23)= 3.49, P < 0.002). To
determine whether this was a result of smaller individ-
ual blinks, fewer trials with blinks, or some combination
of these, data analyses of the single-trial blink frequency,
amplitude and latency were performed. As can be seen
in the top of Fig. 2~ the elderly had an average of 3
trials with startle blinks greater than + 100 tzV and the
young had an average of 24 (t (23) = 3.314, P < 0.005).
When the window was moved to include blinks occur-
ring between 150 and 250 msec post stimulus, there
were very few (average= 1) trials with startle blink
responses in either group, with no difference between
the groups, t (23) = - 0.88.
Because the elderly had so few startle blink re-
sponses, analysis of their amplitudes and latencies may
be unreliable. Nevertheless, the amplitude and latency
of each individual startle blink was measured and a
mean calculated for each subject. Two subjects, one
young and one old, had no startle blinks and were
dropped from this particular blink amplitude and
latency analysis, which included 13 young and 10 old
subjects. Although the average VEOG following the
loud noise burst was much smaller in the elderly than in
the young (Fig. la), individual startle blinks were actu-
ally larger in the old (221 ~V) than in the young (180
/.tV) subjects, t (21) = 2.189, P < 0.05. However. the
startle blink response latency was later in the old (119
msec) than in the young (103 msec) subjects, t (21)=
-2,274, P < 0.05.
To determine whether the reduced number of the
ERPs AND EYEBLINK RESPONSES IN AUTOMATIC AND CONTROLLED PR()CESS1NQ 365
I 't ~ ~ 1 ~H t H t
t -~ + ~t t++l+~
0 40 40
T+lofl EL6 ~'.20~V EO0
Young N= 14
• Old N= 11
Fig. la. Grand average wave forms from 17 scalp electrode sites with non-cephalic reference, recorded during the automatic paradigm.
startle blinks reflected a general reduction in blinking
frequency, the number of spontaneous blinks which
occurred following the presentation of the frequent
background tone was counted. The numbers of trials
with spontaneous blinks exceeding + 100 #V was not
different for the two groups (t (23) = -0.915, n.s.). This
count is displayed in Fig. 2.
Because the elderly seemed to have fewer startle
blinks than the young, we questioned whether this lack
of responsivity to the loud noises might be due to
hearing deficits in the elderly. Hearing deficit values
were calculated by averaging dB hearing levels at 250,
1000, 2000, 4000, and 8000 Hz, for the better ear. These
data were available from 7 elderly subjects whose hear-
ing loss ranged from 14 to 38 dB HL (mean = 27 dB
HL). The correlations between an individual's hearing
deficit and startle blink amplitude and blink frequency
were computed, but neither the amplitude (r = -0.139,
n.s.) nor the number of startle blinks (r = -0.185, n.s.)
was significantly correlated with hearing loss. In ad-
dition, there was no significant correlation between the
number (r=-0.040, n.s.) and amplitude (r=0,308,
n.s.) of startle blinks and the intensity-sensitive N1
amplitude to the frequent tone in the automatic para-
digm for the 10 elderly subjects who had startle blinks.
In Fig, 3 are presented the VEOG peak amplitudes
to the loud noise as a function of trial number. As
stated above, these data do not necessarily reflect blinks.
Nonetheless, in the data from the young subjects, enough
of the trials contained actual startle blinks to show a
strong response decrement over time (r = -0.648, df=
58, P < 0.01), as might be expected for a startle blink
response. In the elderly, there is also a significant VEOG
response decrement over trials (r=-0.382,
P < 0.01), even though, on average, very few trials con-
tained startle blinks. VEOG peak amplitudes following
the frequent tone in this paradigm were not signifi-
cantly correlated with trial number (i.e., time) either for
the young (r= 0.082, n.s.) or for the old (r=0.026,
Conventional ERP analyses
N1 at Cz.
stimuli and was not measured. N1 was measured at Cz
to the frequent event in the automatic paradigm and to
both rare and frequent events in the auditory RT para-
N1 was small in response to the visual
366 J.M. FORD. A. PFEFFERBAUM
Auditory RT Paradigm
Veog • .
0 ~0 800 0 @ 800
~+IO~V EEG ~20uV EOG
Young N= 13
Old N= 1 l
Fig. lb. Grand average wave forms from 17 scalp electrode sites with non-cephalic reference, recorded during the auditory RT paradigm.
digm. The frequent tones in the automatic and auditory
RT paradigms were identical and elicited quite similar
Nls, as can be seen in Table I. Nls to the frequent
tones in the auditory RT paradigm were affected by age
group, being larger (more negative) in the old subjects
( F (1, 22) = 5.02, P < 0.05). A similar size effect in the
automatic paradigm failed to reach statistical signifi-
cance (F (1, 23)= 2.56, P= 0.123). Nls to the rare
tones in the auditory RT paradigm also appeared larger
in the elderly, but again the age effect failed to reach
statistical significance (F (1, 22) = 4.04, P < 0.06).
N1 latencies to the frequent tones were not signifi-
cantly increased in the elderly in the automatic para-
digm (F (1, 23)= 0.025, n.s.) or in the auditory RT
paradigm (F (1, 22) = -0.03, n.s.), nor was N1 latency
increased in the elderly to the rare tone in the auditory
RT paradigm (F (1, 22) = 1.04, n.s.).
P2 at Fz.
In the grand averages, P2 at Fz appeared
larger for older subjects in the auditory RT paradigm
but this was only significant for the frequent tones (F
(1, 22)=4.02, P < 0.05). On the other hand, it ap-
peared larger in the young subjects in the visual RT
paradigm, however this was not statistically significant
for either the rare (F (1, 24) - 1.17, n.s.) or the frequent
event (F (1, 24) = 0.64, n.s.). The mean P2 amplitudes
are presented in Table I.
P2 to the frequent tone in the automatic paradigm
was delayed in the elderly (F (1, 23) = 6.05, P < 0.05).
While the age effect on P2 latency to frequent tones in
the auditory RT paradigm was similar, it was not statis-
N2 at Fz.
N2 is not easily observed in the auditory
ERPs, except at Fz for the young subjects. It was.
however, quite apparent in the visual ERP to the rare
event as can be seen in Fig. lc. Because its amplitude
was affected by the P3 component, it was measured
peak-to-peak (P2-N2) rather than baseline-to-peak. Vis-
ual P2-N2 to the rare event was smaller in the old than
in the young subjects (F (1, 24) = 4.49, P < 0.05). There
were no group differences for the P2-N2 amplitudes to
the frequent visual event (F (1, 24) = 1.99, P < 0.20), to
the rare event in the auditory RT paradigm (F (1, 22) =
0.1, n.s.), to the frequent event in the auditory RT
paradigm (F (1, 22)=0.52, n.s.), or to the frequent
auditory event in the automatic paradigm (F (1, 23) =
0.935, n.s.). The mean values are presented in Table I.
P3 at Fz, Cz, and Pz.
amplitude for each paradigm over the 3 midline elec-
Separate ANOVAs of P3
ERPs AND EYEBLINK RESPONSES IN AUTOMATIC AND CONTROLLED PROCESSING
Visual RT Paradigm
0 ~0 ~0
I i I
T+lo: EE~ Z+.20: EOC
Young N= 15
P4 T6 \ ....... ~
0 ~ 800 0 400
Fig. lc. Grand average wave forms from 17 scalp electrode sites with non-cephalic reference, recorded during the visual RT paradigm.
trodes (Fz, Cz, and Pz) revealed a main effect of age
only for the automatic P3 (F (1, 23)= 20.79, P <
0.0001). However, when only Pz amplitudes were con-
sidered, P3 was significantly smaller in the elderly than
in the young for the automatic (9.4 /zV vs. 25.8 tW; F
(1, 23) = 28.62, P < 0.0001), the auditory RT paradigm
(16.7/~V vs. 22.5 /W; F (1, 22) = 4.98, P < 0.05), and
the visual paradigm (21.2 ttV vs. 26.9/zV; F (1, 24) =
4.58, P < 0.05). These effects are plotted in Fig. 4.
An ANOVA comparing P3 amplitude at Pz across all
paradigms from the 9 old and 13 young subjects, who
had sufficient artifact-free data on all 3 paradigms,
revealed a significant group × paradigm interaction (F
(2, 40)= 4.45, P < 0.05, GG), reflecting the fact that
N1, P2 and P2-N2 amplitudes and S.D.s (in parentheses) in #V.
Automatic Auditory RT Visual RT
Rare Freq. Rare Freq. Rare
* - 3.79 (2.2)
- 5.23 (2.2)
- 4.81 (3.7)
- 7.50 (2.6)
- 3.69 (1.7)
- 5.32 (1.8)
* Missing entries indicate that those components were not measured.
.I.M. FORD, A. PFEFFERBAUM
m YO, JN6
STARTLE BLINK AMPLITUDE
STARTLE BLINK LATENCY
Fig. 2. Top: numbers of startle and spontaneous blinks exceeding
+ 100 #V in young and old subjects. VEOG excursions were counted
as startle blinks if they exceeded +100 /~V, within 50-150 msec
following the loud noise burst. VEOG excursions were counted as
spontaneous blinks if they exceeded + 100 #V within 0-1000 msec
following the frequent background tone. Middle: amplitude of the
startle blinks that met the above criteria. Bottom: latency of the
startle blinks that met the criteria.
the different paradigms were differentially sensitive to
age, with the automatic being the most sensitive, A
similar analysis comparing P3 latency at Pz revealed a
main effect of age on P3 latency (F (1, 20)= 25.18,
P < 0.0001, GG), but not a group × paradigm interac-
tion (F (2, 40)= 1.60, P = 0.22, GG), suggesting that
P3 latency recorded in all 3 paradigms equally reflected
the increase in P3 latency with age.
As can be seen in Fig. 5. in the automatic paradigm
the latency of P3 at Pz was increased in the elderly (338
msec) compared to the young (295 msec) by 43 msec in
the automatic paradigm (F (1, 23)= 4.52, P < 0.05).
The difference was 56 msec in the visual RT paradigm
with the elderly P3 latency at 463 msec and the young
at 407 msec (F (1, 24)= 21.07, P < 0.0001). The dif-
ference was 46 msec in the auditory RT paradigm with
the elderly P3 latency at 364 msec and the young at 318
msec (F (1, 22) = 5.86, P < 0.05).
ANOVA of P3 across the 3 midline sites also con-
firmed a relatively flatter distribution of P3 in the older
subjects compared to that of the younger subjects. This
is apparent in Fig. 4 where the P3 amplitudes from the
3 midline sites are plotted and was reflected in a signifi-
cant group × lead interaction for the automatic (F
(2,46)=13.38, P<0.0005, GG), auditory RT (F
• i • , , i , i , , , i •
1 0 20 30 40 50 60
RARE TRIAL NUMBER
1 O0 200
FREQUENT TRIAL NUMBER
i 200 •
0 100 200
FREQUENT TRIAL NUMBER
Fig. 3. VEOG amplitude decrement over the course of 60 presenta-
tions of the loud noise (top) and 240 presentations of the frequent
ERPs AND EYEBLINK RESPONSES IN AUTOMATIC AND CONTROLLED PROCESSING 369
Fz Cz Pz
AUDITORY RT PARDIGM
Fz Cz Pz
VISUAL RT PARADIGM
Fz Cz Pz
Fig. 4. P3 amplitudes (and S.D.s) measured at peak maximum at Fz,
Cz and Pz.
(2, 44) = 4.26, P < 0.05, GG) and visual RT (F (2, 48)
- 7.16, P < 0.005, GG) paradigms. When P3 was mea-
sured at its maximum value at each of the 3 midline
leads (i.e., not at the Pz-P3 latency), the results are very
similar for the automatic (F (2, 46) = 17.51, P < 0.0005,
GG), auditory RT ( F (2, 44) = 3.85, P < 0.05, GG) and
visual RT paradigm (F (2, 48) = 6.77, P < 0.01, GG).
In the visual RT paradigm, the group × lead interaction
can be attributed to the fact that P3 is larger in the
elderly than the young at Fz and larger in the young
than in the elderly at Pz.
N4 at Fz.
In the automatic paradigm, the N4 peak-
ing at about 400 msec in the young was significantly
larger than in the elderly ( F (1, 23) = 14.95, P < 0.001).
Auditory RT Visual RT
Fig. 5. P3 latencies (and S.D.s) measured at maximum peak amplitude
potential at FP1 and FP2 was considerably smaller in
the elderly subjects, but only significantly so when
measured in the visual RT paradigm (F (1, 24) = 21.35,
P < 0.0001). This decreased negativity and/or enhanced
positivity in the elderly at FP1 and FP2 was also
evident in P3 amplitudes, which were significantly larger
in the old than in the young for the auditory RT
paradigm ( F (1, 22) = 9.04 P < 0.01) and the visual RT
paradigm (F (1, 24 = 14.31, P < 0.001), but not in the
automatic paradigm at FP1 and FP2 ( F (1, 23) = 3.17,
# < 0 10).
Fig. lc illustrates that the slow negative
Scalp distribution analyses
Inspection of the grand average wave forms in Fig. 1
suggests a more uniform scalp distribution in the elderly
than in the young. To determine how variable the P3
amplitude was across the 17 leads, we calculated the
across electrode root mean squared deviation or RMS
(Lehmann and Skrandies 1980) for each subject. The
average RMS for the 2 groups and 3 paradigms are
plotted in Fig. 6a and b. The RMS at the latency of P3
at Pz was significantly smaller in the elderly than in the
young for the automatic (F (1, 23) = 42.37, P < 0.0001),
auditory RT (F (1, 22) = 11).21, P < 0.005) and visual
RT paradigms (F (1, 24) = 14.27, P < 0.001). A similar
analysis was done on the RMS at the latency of N1 at
Cz to the frequent tones in both the automatic and
auditory RT paradigms. Although the old had more
scalp distributional difference than the young, the dif-
ferences were not significant for either the automatic
paradigm (P < 0.27) or for the auditory RT paradigm
(P < 0.10). The variability of N1 amplitude across the
scalp can be appreciated in Fig. 6a for the rare and in
Fig. 6b for the frequent stinmli.
This same anterior/posterior effect was seen when
P3 amplitude measured at the centermost 15 electrode
sites was analyzed for the laterality and the anterior/
posterior factors. There were significant interactions
between group and anterior/' posterior locations for each
Auditory RT Paradigm
J,M. FORD, A. PFEFFERBAUM
Visual RT Paradigm
0 < <0
0 400 800 0 400 800 0 400 800
ms~ msec msec
' ' ' I ' ' ' I ' ' '
400 800 0 400 800 0 400
+10 v. cAv +2 v. RMs
Fig. 6a. The common average (CAVG) with non-cephalic reference across 17 scalp electrode sites (top) and the root mean squared (RMS) voltage
deviation grand average wave forms (bottom) to the rare events, for each paradigm and each group separately. Note the different voltage
calibrations for the different tracings.
paradigm. Amplitudes were normalized to enable a
comparison of scalp distribution differences unaffected
by absolute differences in amplitude across groups.
Analyses revealed significant group xanterior/pos-
terior x laterality interactions for the auditory RT (F
(8, 176) = 2.49, P < 0.05, GG), the automatic (F
(8, 184) = 5.50, P < 0.0005, GG), but not for the visual
RT paradigm (F (8, 192)= 1.70, P=0.16, GG). In-
spection of the means plotted in Fig. 7 suggested post-
hoc pairwise comparisons of P3 recorded from the
mid-frontal locations. For the auditory RT paradigm,
the distribution was flat in the young subjects and
right-maximal in the old. The difference between F3
and F4 was smaller in the young than in the old
subjects (t (22)= 2.07, P=0.05). For the automatic
paradigm, the mid-frontal distribution was again flat
for the young subjects and centrally maximal for the
old. The difference between P3 recorded at Fz and the
electrodes immediately adjacent (F3 and F4) was greater
in the elderly than the young for Fz-F3 (t (23) = 5.00,
P < 0.0001) and for Fz-F4 (t (23) = 2.97, P < 0.01).
Single-trial latency adjustment analyses
The signal/noise screening of the single trials and
the subsequent averaging of those trials yielded 2 types
of ERP comprised of identical trials: 'stimulus-synchro-
nized' and 'latency-adjusted.' To examine the effects of
single-trial latency dispersal on the P3 amplitude, these
two ERPs were compared. (Note: the conventional
stimulus-synchronized averages were made up of ad-
ditional trials that did not pass the signal/noise screen.)
Fig. 8 presents latency-adjusted and stimulus-syn-
chronized averages arbitrarily placed with the P3 peak
at the group median P3 latency. Fig. 8 illustrates that
latency-adjusted P3s were larger than stimulus-synchro-
nized P3s in both the visual RT paradigm (F (1, 24) =
97.06, P < 0.0001, GG) and the auditory RT paradigm
(F (1, 22) = 32.49, P < 0.0001, GG). Although the
young subjects' P3s appeared to be enhanced by the
latency adjustment procedure more than the elderly
subjects' P3s, the age x analysis method interaction was
not significant in either the visual RT paradigm (F
(1, 24) = 3.12, P < 0.10) or the auditory RT paradigm
ERPs AND EYEBLINK RESPONSES IN AUTOMATIC AND CONTROLLED PROCESSING
Automatic Paradigm Auditory RT Paradigm
Visual RT Paradigm
0 ~ -- 0
< < ,,
0 400 800 0 400 800
0 400 800
5~V CAVG ~
i t t i a J i i i
Fig. 6b. The common average (CAVG) with non-cephalic reference across 17 scalp electrode sites (top) and the root mean squared (RMS) voltage
deviation grand average wave forms (bottom) to the frequent events, for each paradigm and each group separately. Note the different voltage
calibrations for the different tracings.
( F (1, 22) = 2.03, P < 0.20). Following the latency ad-
justment procedure, there was still a significant effect of
age on P3 amplitude in both the auditory RT paradigm
('F (1, 22) = 4.88, P < 0.05) and the visual RT paradigm
(F (1, 24)= 4.76, P < 0.05). This is best seen in the
overlays of young and old latency-adjusted ERPs in
Fig. 9, and it suggests that small P3s in the elderly are
not simply due to greater single-trial latency variability.
In fact, when the standard deviations of the single-trial
P3 latencies were compared, there were no significant
differences between young and old subjects for either
the auditory RT paradigm (t (22)= -0.3, n.s.) or the
visual RT paradigm (t (24)= -0.37, n.s.).
The numbers of trials passing the signal/noise screen
did not differ across age groups in either paradigm. In
the auditory RT paradigm, the elderly dropped from an
average of 41 trials included in the conventional ERP
wave forms to 34 in the signal/noise screened trials; the
young dropped from 46 to 39. While there was a main
effect of analysis method for the auditory RT paradigm
(F (1, 22) = 68.82, P < 0.0001, GG), there was no effect
of age (F (1, 22) = 1.37, n.s.), nor was there an age x
analysis method interaction (F (1, 22) = 0.021, n.s.). In
the visual RT paradigm, the elderly dropped from an
average of 47 trials included in the conventional ERP to
41 in the screened ERPs; the young dropped from 41 to
35. Again, there was a main effect of analysis method
(F (1, 24) = 66.8, P < 0.0001, GG), although there was
no effect of age (F (1, 24) = 1.67, n.s.) nor was there an
interaction of age × analysis method (F (1, 24) = 0.07,
The signal/noise screening procedure also yielded
cross-products and correlation coefficients for each trial.
They indicated the goodness of fit between points of the
P3 epoch and template in terms of amplitude and
shape, respectively. The cross-products were larger for
the young than old, as would be expected from their
larger P3 amplitudes for both the auditory RT para-
digm (t (22) = 2.98, P < 0.01) and the visual RT para-
digm (t (24)= 2.48, P < 0.05). The shape of the tem-
plate (as measured by the correlation coefficients) did
not fit better for the young than for the old in either the
i i i
3 4 5
J.M. FORD, A. PFEFFERBAUM
= i i
2 4 5
AUDITORY RT PARADIGM
! i i i
2 3 4 5 -0.2
AUDITORY RT PARADIGM
VISUAL RT PARADIGM 1.2"
, i , , -0.2
1 2 3 4 5
VISUAL RT PARADIGM
3 ; ;
Fig. 7. Normalized P3 amplitudes for the 3 × 5 electrode montage, going from left-most to right-most scalp site for each of the frontal, central, and
parietal rows, for the automatic, auditory RT, and visual RT paradigms.
auditory RT paradigm (t (22)= 1.87, P < 0.08) or the
visual RT paradigm (t (24) = 0.86, n.s.).
The median P3 latency of the single trials was calcu-
lated for each subject and compared across age groups.
As was true for the conventional average P3, the median
P3 latency was later in the elderly than in the young
ERPs AND EYEBLINK RESPONSES IN AUTOMATIC AND CONTROLLED PROCESSING 373
LATENCY ADJUSTED vs STIMULUS SYNCHRONIZED
0 400 800 msec
800 400 msec
400 800 msec 0 400 800 msec
Fig. 8. Latency-adjusted and stimulus-synchronized ERPs are overlaid for young (top) and old (bottom) subjects at the group median P3 latency.
subjects for both the auditory RT paradigm (t (22)=
2.72, P < 0.05) and visual RT paradigm (t (24) = 3.34,
P < 0.01).
Average RTs were calculated for trials included in
the conventional averages and the small number of
trials which failed to pass the signal/noise screen (Ta-
ble II). Regardless of method used for calculating the
RTs, the elderly responded about 40-50 msec later than
the young in the auditory RT paradigm, and only about
20 msec later than the young in the visual RT paradigm.
None of the t tests comparing young to old RTs were
significant at the P < 0.05 level. T tests comparing the
methods of calculating RTs were not significant.
The correlation between single-trial P3 latencies and
RTs was calculated for each subject. The coefficients
were z-transformed and compared across age groups.
Average reaction times (msec) from trials passing various single-trial
screening procedures and S.D.s.
procedure Auditory RT Visual RT Auditory RT Visual RT
A: trials without artifacts and errors; B: trials with identifiable P3,
but without artifacts and errors; C: trials without identifiable P3 and
without artifacts and errors.
374 J.M. FORD, A. PFEFFERBAUM
YOUNG vs OLD
-t -I i I :-Jill÷
0 800 msec
i iq-i j i
Young N= 13 Young N= 15
Old N= 11 Old N= 11
Fig. 9. Young and old subjects' latency-adjusted ERPs are overlaid at the group median P3 latency.
There was not a significant age effect for either the
auditory RT paradigm (t (22)=-0.68,
visual RT paradigm (t (24) = 0.35, n.s.). In both groups,
all but one subject had positive P3/RT correlation
coefficients, with the majority being statistically signifi-
n.s.) or the
Both young and old were accurate responders. Both
groups missed an average of less than 1 out of 60 trials
with no significant differences between the groups in
either the auditory RT (t (22) = 1.28, n.s.) or visual RT
(t (24)= 0.89, n.s.) paradigm.
The amplitude of P3 at Pz was reduced in the elderly,
especially in the automatic paradigm. The possibility
that the large positive startle blink of the young subjects
might have augmented their P3 amplitude cannot be
ruled out in spite of the selective deletion of trials
containing blinks (VEOG > 100 gV) in the P3 epoch,
275 600 msec. P3s uncontaminated by eye blink are
seen at FP1 and FP2 in Fig. la. Although at frontal
sites the task-dependent P3 can actually be larger in the
elderly than in the young (Mullis et al. 1985), the
automatic P3 is smaller in the elderly, even at the
Another measure of automatic responsiveness might
be the startle blink, which was often missing in the
elderly. The fact that both the startle and spontaneous
blinks were at least as large, if not larger, in the old as
in the young is consistent with the literature (Tecce et
al. 1980) and suggests that the fewer blinks in the
elderly to the startling stimulus are not due to a general
decline in the ability to blink. Instead, we suggest that
the fewer and delayed startle blinks in the elderly may
reflect an automatic processing difference, perhaps re-
flecting reduced reactivity. This interpretation must be
tempered by the possibility that presbycusis may have
rendered the loud noise bursts less startling, even though
auditory thresholds did not correlate with startle blink
frequency within the elderly group.
The possibility that startle blinks are responsible for
the automatic P3 or that they are a necessary step in the
pathway leading to the generation of the automatic P3
was addressed in a post-hoc analysis. By separately
averaging trials with and without startle blinks (> 100
/~V) across subjects, we found that in young subjects, P3
is large regardless of whether it was accompanied by a
startle blink (23.4 /~V) or not (24.1 gV). In the elderly,
this comparison is more difficult to make because there
are so many fewer trials with startle blinks than without
startle blinks. However, VEOG amplitude is not signifi-
cantly correlated with P3 amplitude in young (r=
-0.326, P=0.25, df=12) or in the old (r=0.368,
P = 0.26, dr= 9) subjects. These data suggest that the
automatic P3 amplitude does not depend upon the
presence of a large startle blink, per se.
While there are many studies showing age-related P3
amplitude reduction in paradigms involving tasks
(Marsh 1975; Goodin et al. 1978; Harbin et al. 1984:
Picton et al. 1984: Mullis et al. 1985; Pfefferbaum and
ERPs AND EYEB[.INK RESPONSES IN AUTOMATIC AND CONTROLLED PROCESSING 375
Ford 1988), fewer have attempted to record P3s from
the elderly in tasks eliciting automatic processing.
Knight (1987) recorded P3s to novel stimuli (simulated
dog barks) embedded in a series of target and non-target
tones. His finding of reduced and delayed P3s to the
unattended novel stimuli in the elderly is consistent
with our data and suggests that the capacity for this
type of automatic processing declines with age.
This interpretation is not fully consistent with behav-
ioral studies showing that the elderly are only deficient
in controlled processing (Plude and Hoyer 1981; Rab-
bitt 1981). It is also not consistent with an ERP task
used earlier in this laboratory in which the pattern of P3
amplitudes was interpreted as reflecting the automatic
process of extracting probability information from a
sequence. In that task, we found that neither the pattern
of P3 amplitudes nor the overall P3 amplitude differed
between young and old subjects (Ford et al. 1982a).
Perhaps the startling stimuli used in the current study
evoke a qualitatively different type of automatic
processing from that invoked by the unnecessary and
task-irrelevant tracking of event frequency and prob-
ability in the earlier study. The reflexive contraction of
middle ear muscle(s) to loud stimuli (the acoustic reflex)
is also reduced in the elderly (Hall 1982). Startling
stimuli evoke an extremely primitive low-order auto-
matic process. While it is quite possible that deficits on
the sensory side of the startle reflex loop may explain
the fewer number of startle responses in the elderly, it is
also possible that the functioning of a primitive auto-
matic process is delayed and less reliable in the elderly.
We must also remain open to the possibility that the
elderly actively inhibit their response to the loud noise
The psychological meaning of the task-dependent P3
recorded at Pz has been the subject of lively debate
recently (Donchin and Coles 1988; Verleger 1988). The
Verleger model proposes that P3 amplitude decreases
with increased task difficulty, but increases with in-
creased effort, The elderly subjects' button press re-
sponses were not different from the young and thus
they responded more quickly and accurately than ex-
pected for their age. Therefore, it is unlikely that their
small P3s at Pz reflect the expenditure of less effort.
The smaller P3s at Pz might have resulted from the
tasks being more difficult for the elderly. However,
according to the Verleger model, one would expect that
equivalent performance in a more difficult task would
require more effort and, hence, a larger P3, instead of
the smaller P3 reported here. Unless the effort factor in
the Verleger model receives a considerably smaller
weighting than the difficulty factor, this model is not
useful in describing the psychological meaning of the
small P3 at Pz in the elderly. The Donchin and Coles
model suggests that small Pz P3s in the elderly could
reflect deficiencies in updating their event expectancies,
but the performance data suggest that the current task
was not sufficiently difficult to test this model (Donchin
and Coles 1988). A third possibility is that P3 amplitude
decrements at Pz in the elderly are not directly related
to psychological variables at all, but are secondary to
the ubiquitous, yet functionally ill-defined, anatomical
brain changes accompanying normal aging.
Structural brain changes in the elderly might also
underlie the considerable uniformity of P3 amplitude
across the 17 scalp sites. However, any attempt to
attribute this lack of topographic variation in P3 to
uniform changes in skull thickness or global changes in
brain morphology (e.g., age-related global atrophy) must
first account for the N1 RMS results which show that
N1 was not more variable in the young than in the old.
To account for both the P3 and N1 distribution find-
ings, different age-related changes in N1 and P3 genera-
tors might be postulated, or regionally specific changes
in skull, scalp, or CSF (Pfefferbaum 1990) could also be
considered. The contribution of the non-cephalic refer-
ence to these observations must also be considered.
We~have not previously found P3 to be significantly
reduced with age, either to auditory (Ford et al. 1979a,b;
Pfefferbaum et al. 1980a) or visual stimuli (Ford et al.
1979b, 1982b). Rather, we have found that the change
in P3 amplitude with age is manifest primarily as a
change in topography and have suggested that it is due
to a decreased frontal slow wave negativity. Our present
finding of decreased frontal negativity extends our find-
ings (Pfefferbaum et al. 1980b), and those of Picton et
al. (1984), to include more frontal and lateral scalp sites
than previously observed.
This lack of frontal negativity may contribute to the
large frontal P3 in the elderly, especially pronounced in
the visual paradigm. Large frontal P3s to visual stimuli
are consistent with the data of Mullis et al. (1985) who
reported that the visual P3 is larger in the elderly than
in the young at the frontal electrodes and smaller than
the young P3 at the parietal sites. Inspection of the
grand average wave forms of Looren de Jong et al.
(1988) indicates a similar pattern of larger visual P3s at
frontal electrode sites in the elderly. Lack of frontal
negativity in the elderly is also reflected in their small
N4 components observed in the automatic paradigm.
Confirming our previous work, we found that adjust-
ing the ERP on P3 latency at Pz did not change the
effect of age on P3 amplitude (Pfefferbaum et al.
1980a,b). These analyses suggest that individual P3s
recorded at Pz in the elderly are smaller than those of
the young. Confirming other reports, we found that
single-trial P3 latencies are not more variable in the old
subjects than in the young and that the correlation of
single-trial P3 latencies with RTs is not smaller in the
elderly than in the young (Pfefferbaum et al. 1980a,b;
Patterson et al. 1988).
The finding of P3 amplitude laterality differences
376 J.M. FORD, A. PFEFFERBAUM
between the young and old subjects was not expected.
There seemed to be a trend for the elderly to have
right-sided maxima at the frontal locations in the reac-
tion time paradigms. In a counting task, requiring no
button press, Picton et al. (1984) did not observe any
laterality differences across the age groups studied.
Whether these age-related laterality effects can be repli-
cated and whether they interact with button pressing
requires further investigation.
As we expected, P3 was delayed in the elderly sub-
jects in all paradigms, regardless of how the average was
constructed. A lack of group x paradigm interaction for
P3 latency suggests that the delay was not exaggerated
differentially in the different paradigms. P3 latency
delay in the elderly is consistent with a large body of
literature showing delays to both auditory (Goodin et
al. 1978; Ford et al. 1982a; Syndulko et al. 1982; Brown
et al. 1983; Polich et al. 1983; 1985; Pfefferbaum et al.
1984; Fein and Turetsky 1989) and visual stimuli (Ford
et al. 1979b, 1982a,b; Pfefferbaum et al. 1984).
Both N1 and P2 to the frequent tones in the auditory
RT paradigm were larger in the elderly than in young
subjects. While this is consistent with some of our
previous work (Pfefferbaum et al. 1984), it is in conflict
with other reports of smaller (Goodin et al. 1978) or
unchanged Nls or N1-P2s (Spinkl et al. 1979; Pfeffer-
baum et al. 1980b; Michalewski et al. 1982; Picton et al.
1984; Patterson et al. 1988). In some of our previous
work, we adjusted the auditory stimuli relative to each
subject's threshold, but in this investigation the inten-
sity was the same for all subjects. N1 amplitude is both
sensory and attention dependent and may reflect an
overall higher level of general attention (Hansen and
Hillyard 1980) in the elderly during this auditory RT
task. It may be this higher level of attention that en-
abled the elderly to push the RT button faster than
expected for their age. The use of a non-cephalic refer-
ence may account for some of the amplitude observa-
It is interesting to note that the Nls to the back-
ground tones in the automatic and auditory RT para-
digms were elicited by the same stimuli and are similar
in amplitude for both groups, giving evidence of relia-
bility of N1 amplitude across time and task in both
young and old subjects.
P3 amplitude and latency have been proposed as
possible trait and state markers for a variety of CNS
disorders (dementia, alcoholism, schizophrenia; Pfeffer-
baum et al. 1990). The current study explores normal
age-related changes which must be taken into account
before clinical applications can be pursued. Of particu-
lar note is the profound change in P3 topography seen
in the elderly, making it necessary to qualify topo-
graphically any statements about age-related changes in
Special thanks are extended to Patricia M. White for her help in
computer programming, data processing and statistical analysis, to
Susan Hammen for assistance with the figures, to Cathy Rodgers and
Ann Doty for collecting the data, and to Margaret Rosenbloom for
Brown, W., Marsh, J.T. and LaRue, A. Exponential electrophysiologi-
cal aging: P3 latency. Electroenceph. clin. Neurophysiol., 1983, 55:
Donchin, E. and Coles, M.G.H. Is the P300 component a manifesta-
tion of context updating? Behav. Brain Sci., 1988, 11: 357-374.
Fein, G. and Turetsky, B. P300 latency variability in normal elderly:
effects of paradigms and measurement technique. Electroenceph.
clin. Neurophysiol., 1989, 72: 384-394.
Ford, J.M. and Pfefferbaum, A. Age-related changes in ERPs. In:
P.K. Ackles, J.R. Jennings and M.G. Coles (Eds.), Advances in
Psychophysiology. JAI Press, Greenwich, CT, 1985: 301-339.
Ford, J.M., Hopkins, III, W.F., Roth, W.T., Pfefferbaum, A. and
Kopell, B.S. Age effects on event-related potentials in a selective
attention task. J. Gerontol., 1979a, 34: 388-395.
Ford, J.M., Roth, W.T., Mohs, R.C., Hopkins, 111, W.F. and Kopell,
B.S. The effects of age on event-related potentials in a memory
retrieval task. Electroenceph. clin. Neurophysiol., 1979b, 47: 450-
Ford, J.M., Duncan-Johnson, C.C., Pfefferbaum, A. and Kopell, B.S.
Expectancy for events in old age: stimulus sequence effects on
P300 and reaction time. J. Gerontol., 1982a, 37: 696-704.
Ford J.M., Pfefferbaum, A. and Kopell, B.S. Effects of perceptual and
cognitive difficulty on P3 and RT in young and old adults.
Electroenceph. clin. Neurophysiol., 1982b, 54:311-321.
Geyer, M.A. and Braff, D.L. Startle habituation and sensorimotor
gating in schizophrenia and related animal models. Schizophr.
Bull., 1987, 13: 643-668.
Goodin, D., Squires, K., Henderson, B. and Starr, A. Age-related
variations in evoked potentials to auditory stimuli in normal
human subjects. Electroenceph. clin. Neurophysiol., 1978, 44:
Hall, J.W. Acoustic reflex amplitude. I. Effect of age and sex. Audio-
logy, 1982, 21: 292-309.
Hansen, J.C and Hillyard, S.A. Endogenous brain potentials associ-
ated with selective auditory attention. Electroencepb. clin. Neuro-
physiol., 1980, 49: 277-290.
Harbin, T., Marsh, G. and Harvey, M. Differences in the late compo-
nents of the event-related potential due to age and to semantic and
non-semantic tasks. Electroenceph. clin. Neurophysiol., 1984, 59:
Hasher, L. and Zacks, R.T. Automatic and effortful processes in
memory. J. Exp. Psychol.: Gen., 1979, 108: 356-388.
Knight, R.T. Aging decreases auditory event-related potentials to
unexpected stimuli in humans. Neurobiol. Aging, 1987, 8: 109-113.
Lehmann, D. and Skrandies, W. Reference-free identification of com-
ponents of checkerboard-evoked multichannel potential fields.
Electroenceph. clin. Neurophysiol., 1980, 48: 609-621.
Looren de Jong, H., Kok, A. and Van Rooy, J.C.G.M. Early and late
selection in young and old adults: an event-related potential study.
Psychophysiology, 1988, 25: 657-671.
Marsh, G.R. Age differences in evoked potential correlates of a
memory scanning process. Exp. Aging Res., 1975, 1: 3-16.
McCarthy, G. and Wood, C.C. Scalp distributions of event-related
potentials: an ambiguity associated with analysis of variance mod-
els. Electroenceph. clin. Neurophysiol., 1985, 62: 203-208.
ERPs AND EYEBLINK RESPONSES 1N AUTOMATIC AND CONTROLLED PROCESSING 377 Download full-text
Michalewski, H.J., Patterson, J.V., Bowman, T.T., Litzelman, D. and
Thompson, L.W. A comparison of the emitted late positive poten-
tial in older and young adults. J. Gerontol., 1982, 37: 52-58.
Mullis, R.J., Holcomb, P.J., Diner, B.C. and Dykman, R.A. The
effects of aging on the P3 component of the visual event-related
potential. Electroenceph. clin. Neurophysiol., 1985, 62: 141-149.
N~iatfinen, R. and Picton, T, The N1 wave of the human electric and
magnetic response to sound: a review and an analysis of the
component structure. Psychophysiology, 1987, 24:375 425.
Ohman, A. The orienting response, attention and learning: an infor-
mation processing perspective. In: H.D. Kimmel, E.H. Van Olst
and J.F. Orlebeke (Eds.), The Orienting Reflex in Humans.
Lawrence Erlbaum, Hillsdale, N J, 1979: 443-471.
Patterson, J.V., Michalewski, J.J. and Starr, A. Latency variability of
the components of the auditory event-related potentials to infre-
quent stimuli in aging, Alzheimer-type dementia and depression.
Electroenceph. clin. Neurophysiol., 1988, 71:450 460.
Pfefferbaum, A. Model estimates of CSF and skull influences on
scalp-recorded ERPs. Alcohol, 1990, 7: 479-482.
Pfefferbaum, A. and Ford J.M. ERPs to stimuli requiring response
production and inhibition: effects of age, probability and visual
noise. Electroenceph. olin. Neurophysiol., 1988, 71: 55-63.
Pfefferbaum, A., Ford J.M., Roth, W.T. and Kopell, B.S. Age dif-
ferences in P3-reaction time associations. Electroenceph. clin.
Neurophysiol., 1980a, 49:257 265.
Pfefferbaum, A., Ford, J.M., Roth, W.T. and Kopell, B.S. Age-related
changes in auditory event-related potentials. Electroenceph. clin.
Neurophysiol., 1980b, 49: 266-276.
Pfefferbaum, A., Ford, J.M., Wenegrat, B., Roth, W.T. and Kopell,
B.S. Clinical application of the P3 component of event-related
potentials. I. Normal aging. Electroenceph. clin. Neurophysiol.,
1984, 59: 85-103.
Pfefferbaum, A., Ford, J.M., White, P. and Roth, W.T. P3 in schi-
zophrenia is affected by stimulus modality, response requirements,
medication status and negative symptoms. Arch. Gen. Psychiat.,
1989, 46: 1035-1046.
Pfefferbaum, A., Ford. J.M. and Kramer, H.C. Clinical utility of long
latency cognitive event-related potential: the cons. Electroenceph.
clin. Neurophysiol., 1990, 76: 6-12.
Picton, T.W., Stuss, D.T., Champagne, S.C. and Nelson, R.F. The
effects of age on human event-related potentials. Psychophysi-
ology, 1984, 21: 312-325.
Plude, D.J. and Hoyer, W.J. Adult age differences in visual search as
a function of stimulus mapping and processing load. J. Gerontol.,
1981, 36: 598-604.
Polich, J., Howard, L. and Starr, A. P300 latency correlates with digit
span. Psychophysiology, 1983, 20: 665-669.
Polich, J., Howard, L. and Starr, A. Aging effects on the P300
component of the event-related potential from auditory stimuli:
peak definition, variation and measurement. J. Gerontol.. 1985,
Rabbitt, P.M.A. Cognitive psychology needs models for changes in
performance with old age. In: J. Long and A. Baddeley (Eds.),
Attention and Performance IX. Erlbaum, Hillsdale, NJ, 1981:
Roth, W.T., Dorato, K,H. and Kopell, B.S. Intensity and task effects
on evoked physiological responses to noise bursts. Psychophysi-
ology, 1984, 21: 466-481.
Ruchkin, D. and Glaser, E. Simple digital filters for examining CNV
and P300 on a single-trial basis. In: D.A. Otto (Ed.), Multidisci-
plinary Perspectives in Event-Related Brain Potential Research.
U.S. Environmental Protection Agency, Washington, DC, 1978:
Spinkl, U., Johannsen, H.S. and Pirsig, W. Acoustically evoked poten-
tial dependent upon age. Scand. Audiol., 1979, 8:11 14.
Stephenson, W.A. and Gibbs, F.A. A balanced non-cephalic reference
electrode. Electroenceph. clin. Neurophysiol., 1951, 3:237 240.
Syndulko, K., Hansch, E.C., Cohen, S.H., Pearce, J.W., Goldberg, Z.,
Montan, B., Tourtellotte, W.W. and Potvin, AR. Long-latency
event-related potentials in normal aging and dementia. In: J.
Courjon, F. Maugui+re and M. Revol (Eds.), Clinical Applications
of Evoked Potentials in Neurology. Raven Press, New York, 1982:
Tecce, J.J., Yrachik, D.A., Meinbresse, D., Dessonville, C.L. and Cole,
J.O. CNV rebound and aging. I. Attention functions. In: Motiva-
tion, Motor and Sensory Processes of the Brain. Progr. in Brain
Res., Vol. 54. Elsevier, Amsterdam, 1980:553 561.
Verleger, R. Event-related potentials and cognition: a critique of the
context updating hypothesis and an alternative interpretation of
P3. Behav. Brain Sci., 1988, 11: 343-356.